| 张圆圆,邓文礼,田茂再.基于变系数模型的自适应分位回归方法[J].数学年刊A辑,2012,33(5):539~556 |
| 基于变系数模型的自适应分位回归方法 |
| Adaptive Quantile Regression Based on Varying-CoefRcient Models |
| |
| DOI： |
| 中文关键词: 自适应加权光滑 条件分位 齐性检验 局部多项式 变系数 |
| 英文关键词:Adaptive weights smoothing, Conditional quantile, Homogeneity
test, Local polynomial, Varying-Coefficient |
| 基金项目:国家自然科学基金(No11271368);全国统计科研计划项目(No2011LZ031);教育部人文社会重点研究基地重大项目(No08JJD910247);教育部科学技术研究重点项目(No108120);北京市哲学社会科学规划项目(No12JGB051);中国人民大学科学研究基金项目(No10XNK025);中国人民大学科学研究基金项目(重大基础研究计划)(No10XNL018)资助的项目 |
| Author Name | Affiliation | E-mail | | ZHANG Yuanyuan | School of Statistics,Renmin University of China,Beijing 100872,china | yuanzhang@ruc.edu.cn | | TANG Manlai | School of Mathematics,Baptist University of Hong Kong,Kowloon Tong,Hong Kong,china | mltang@math.hkbu.edu.hk | | TIAN Maozai | School of Statistics,Renmin University of China,Beijing 100872,china | mztian@ruc.edu.cn |
|
| Hits: 1429 |
| Download times: 37 |
| 中文摘要: |
| 提出了变系数模型条件分位估计的一种新方法.变系数模型已经成为经济学、流行病学、纵向数据和医学领域处理高维数据的有力工具.该模型有助于探测数据的动态特征、降低模型偏差、避免高维灾难,同时便于解释.尽管关于变系数模型条件均值的估计已经有很多文章,但关于变系数模型条件分位的估计方面的文章相对较少.文中提出了一种有效的适应性分位回归方法来诊断出齐性邻域,进行局部自适应窗宽选择和局部线性逼近,同时给出了估计量的风险界和最优窗宽的自动选择准则.模拟研究说明了所提出估计方法的效果. |
| 英文摘要: |
| This paper proposes a new technique for conducting conditional quantile estima-
tion in varying-coefficient models. Varying-coefficient models have become powerful tools for
processing high-dimensional data in economical, epidemiological, longitudinal and medical
studies. They are useful to explore the dynamic feature, reduce modeling bias, alleviate
the “curse of dimensionality”, and provide easy interpretation. The main contributions of
this paper include the development of a more efficient technique called the locally adaptive
bandwidth (LAB) for determining particular neighborhoods to construct local linear approx-
imations and the establishment of some tight risk bounds for the estimation as well as an
automatic selection principle for the optimal bandwidth. Simulation studies are conducted
to illustrate the performance of the proposed estimates. |
| View Full Text View/Add Comment Download reader |
| Close |
|
|
|
|
|
